Article Text

Download PDFPDF

Could CT scan avoid unnecessary flexible bronchoscopy in children with active pulmonary tuberculosis? A retrospective study
  1. Kévin Arlaud1,
  2. Guillaume Gorincour2,
  3. Julien Bouvenot3,
  4. Hervé Dutau4,
  5. Jean-Christophe Dubus1
  1. 1Department of Pediatrics and CNRS URMITE 6236, University Timone Hospital, Marseille, Franc
  2. 2Department of Pediatric Radiology, University Timone Hospital, Marseille, France
  3. 3Department of Public Health, University Timone Hospital, Marseille, France
  4. 4Thoracic Endoscopy Unit, University Sainte Marguerite Hospital, Marseille, France
  1. Correspondence to Dr Jean-Christophe Dubus, Unité de Médecine Infantile, CHU Timone Enfants, 264 rue Saint-Pierre, 3385 Marseille Cedex 05, France; jean-christophe.dubus{at}


Background Systematic flexible bronchoscopy is advocated in the initial management of childhood pulmonary tuberculosis.

Methods Our aim was to describe and to compare the systematic flexible bronchoscopy findings of 53 children (6.5±4.4 years; 52.8% boys) with active pulmonary tuberculosis to their initial clinical and radiological (chest x-ray, n=53; CT, n=45) features in an 11-year retrospective study.

Results Flexible bronchoscopy was normal in 51% of cases. A severe tracheobronchial involvement (extrinsic compression >50% or obstructive endoluminal mass >25% of the lumen) was found in 10 patients. Flexible bronchoscopy guided therapy in 13 cases (steroids n=12, bronchoscopic extraction of a granuloma n=1) and permitted isolation of Mycobacterium tuberculosis in three patients (5.7%). No useful information was obtained by flexible bronchoscopy in 73.5% of cases. No correlation was found between flexible bronchoscopy findings and clinical features or x-ray findings. Conversely, a strong correlation was found between severe bronchoscopic involvement and tracheobronchial luminal narrowing on CT. The CT negative predictive value was 100% (95% CI 91% to 100%). Based on CT results, flexible bronchoscopy could have been avoided in about 60% of our patients.

Conclusions Flexible bronchoscopy remains a very relevant tool in the diagnostic and therapeutic management of childhood pulmonary tuberculosis but resulted in treatment modification or microbiological proof in a minority of our patients. We propose that flexible bronchoscopy in children with pulmonary tuberculosis be limited to those who show tracheobronchial luminal narrowing on an initial CT scan.

Statistics from


Tuberculosis remains a public health problem in France, particularly in urban areas (incidence: 8.5/100 000 in 2006).1 The isolation of Mycobacterium tuberculosis is a challenge especially in children and microbiological confirmation is obtained in less than 50% of cases.2,,4 Treatment is usually initiated only on high clinical suspicion supported by a tuberculin skin test and chest x-ray results.5

The development and availability of tools such as flexible bronchoscopy and chest CT raise the question of their use in the management of childhood tuberculosis.6,,9 The French national recommendations advocate the systematic use of flexible bronchoscopy for any child with pulmonary tuberculosis.10 However, flexible bronchoscopy is still an invasive procedure which can be associated with morbidity and may require general anaesthesia.11,,15 Flexible bronchoscopy cannot be entirely justified by the low microbiological recovery rate in children, but it is useful for detecting possible tracheobronchial obstruction and guiding steroid therapy or surgical intervention. The use of systematic flexible bronchoscopy in pulmonary tuberculosis is based on previous studies which reported that tracheobronchial involvement was undetectable by clinical features or chest x-ray.16 17 Nevertheless, no study has investigated the role of CT in actually identifying tracheobronchial involvement.

What is already known on this topic

  • Flexible bronchoscopy for detecting endobronchial disease in active childhood tuberculosis is frequently normal.

  • Clinical or chest x-ray features do not select children with endobronchial disease.

What this study adds

  • Positive CT findings performed before flexible bronchoscopy suggest severe endobronchial involvement.

  • Flexible bronchoscopy could be avoided if no tracheobronchial luminal narrowing is found on CT.

The aim of our study is to describe the bronchoscopic findings in 53 children with tuberculosis disease and to compare these results with the data provided by CT.

Methods and materials


We conducted a retrospective study enrolling children for whom a flexible bronchoscopy was performed for tuberculosis disease in our department between 1995 and 2006, our department being the only paediatric bronchoscopy facility for Marseille and the surrounding area, which have a population of nearly 2 million people. During this period, flexible bronchoscopy was routinely performed in children with an abnormal chest x-ray with lymphadenopathy and suspected tuberculosis on a positive tuberculin skin test above 15 mm with a 10 tuberculin unit (TU) Mantoux test, performed because of a contact history and/or clinical feature suggestive of tuberculosis.10 18 19 Treatment was a combination of isoniazid, rifampin and pyrazinamide for 2 months followed by izoniazid and rifampin for 4 months.20

The patient’s medical records were examined for vaccine statute, BCG status, a history of close contact with tuberculosis in an adult and the presence at the time of diagnosis of the following symptoms: cough, fever, anorexia, asthenia, weight loss and nocturnal sweats. Patients were followed until the end of treatment, that is, for 6 or 9 months.

Bronchoscopic findings

Flexible bronchoscopy was performed under general anaesthesia with a laryngeal mask or under local anaesthesia and sedation using intravenous midazolam. Lidocaine was then applied locally with a maximum dose of 4 mg/kg.13 All examinations were performed by one of two operators using either with a Pentax FB-10X or a Pentax FB-15P. Bronchoscopic findings were localised following lung segmentation and classified as normal endoscopy, bronchial mucosal inflammation, granulation tissue, extrinsic tracheobronchial compression of more or less than 50% of the lumen, and obstructive endoluminal mass (polyp or caseum) of more or less than 25% of the lumen. Steroid therapy (prednisone 1–2 mg/kg/day) was indicated when the patient had evidence of severe endoscopic involvement, that is, extrinsic compression of more than 50% of the lumen or an obstructive endoluminal mass of more than 25% of the lumen. Four weeks after this treatment was initiated the dose was tapered off according to the clinical and radiological evolution and in some cases on the results of repeated flexible bronchoscopy.

Bronchial washing in the area of x-ray abnormality was routinely performed by the instillation and aspiration of two aliquots of 5 ml physiologic saline in order to obtain microbiological proof of tuberculosis. The samples were screened for acid-fast bacilli and cultured in adapted medium to isolate M tuberculosis. As other microbiological samples (sputum or gastric lavage) were not obtained using standardised techniques they could not be used for comparison with bronchial washing samples.

Radiological findings

All patients underwent both chest x-ray and CT. Only the anteroposterior view of the chest x-ray was analysed. CT examinations were performed with a single-detector device from 1995 to December 2003 and with a 16-detector device from December 2003 to date. All radiological data were read retrospectively for this study by the same paediatric radiologist blinded to the clinical and bronchoscopic findings. For CT analysis, attention was specifically focused on the following abnormalities and their localisation: parenchymal involvement (area of consolidation, nodule, miliary), hilar or peritracheal lymphadenopathy, tracheobronchial luminal narrowing, atelectasis and pleural effusion. Tracheobronchial luminal narrowing was defined by whether there was extrinsic compression or an obstructive endoluminal mass from the trachea to segmental bronchi. The degree of narrowing was not quantified. We have compared flexible bronchoscopy and CT data only for those patients with a CT performed less than 15 days before flexible bronchoscopy to avoid any modification caused by the flexible bronchoscopy procedure.

Statistical analysis

Data are shown as percentages or means±SD. SPSS v 13.0.1 software was used for statistical analysis. Significant associations between bronchoscopic findings and clinical/radiological findings were studied using Pearson’s χ2 test or Fisher’s exact test for qualitative data and the Mann–Whitney test for quantitative data. Differences between children aged >2 years and ≤2 years were analysed. Children who had or did not have a CT within 15 days before bronchoscopy were compared. CT sensitivity, specificity and predictive values (as well as their 95% CIs) were assessed using bronchoscopic findings as the gold standard. A p value ≤0.05 was considered statistically significant for all tests.


Patients and clinical features

Between 1995 and 2006, we performed flexible bronchoscopy for pulmonary tuberculosis in 67 children. Only 53 of the children (28 boys; 52.8%) had been diagnosed with and treated for tuberculosis disease. Their median age was 5.3 years (range 3 months to 16 years) and 11 patients were aged less than 2 years old (20.8%). Thirty eight patients (71.7%) had been vaccinated with BCG. A history of close contact with tuberculosis in an adult was found in 52% of cases. One third of the children (n=18) were considered to be symptom-free at the time of flexible bronchoscopy. For the others, cough (n=25; 47.2%), fever (n=15; 28.3%), weakness (n=15; 28.3%), anorexia (n=13; 24.5%), weight loss (n=8; 15.1%) and nocturnal sweats (n=2; 3.8%) were found. There was no significant difference between clinical presentation in the children aged 2 or aged less than 2.

Bronchoscopic findings and consequences

Overall, 77% of fiexible bronchoscopy procedures were performed under local anaesthesia and sedation. Tolerance was good apart from an episode of nasal bleeding and an episode of transitory hypoxaemia without bradycardia. No major complications occurred.

Flexible bronchoscopy was entirely normal in 51% of cases. Abnormalities are presented in table 1. The most frequent abnormality was an extrinsic compression of less than 50% of the lumen (n=13; 24.5%). Severe disease as defined earlier was found in 18.9% of the children (extrinsic compression of more than 50%, n=4; obstructive endoluminal mass of more than 25%, n=6). Three of the 10 patients with severe disease were symptom free. No significant link was found between age, clinical features and severity of bronchoscopic involvement.

Table 1

Initial bronchoscopic findings in 53 children with active tuberculosis

Twelve patients (22.6%) were given steroids, 10 of whom had severe bronchoscopic involvement and 2 of whom had disease outside our study criteria (one patient with an extrinsic compression ranging from 30% to 40% and one with a granuloma obstructing the bronchial lumen by about 20%). For these 12 patients a second flexible bronchoscopy was performed between 14 and 77 days (mean 28.5 days) after the first procedure. Bronchoscopic abnormalities had resolved in nine, two had stable lesions and one had worsening bronchial compression. Consequently a third bronchoscopy was performed for these three patients: one showed complete resolution, one partial improvement with a 30% residual reduction in the diameter of the right upper lobar bronchus, and one a persistent granuloma requiring extraction by rigid bronchoscopy. No indication for surgical resection was noted during our study period.

M tuberculosis was isolated from the bronchial washing fluid in only three of the 53 patients (5.7%).

The initial flexible bronchoscopy resulted in no treatment modification or bacteriological proof in 73.5% (39/53) of cases.

Radiological findings and comparison with bronchoscopic findings

Given the inclusion criteria, all the patients had chest x-rays consistent with a diagnosis of active tuberculosis disease (hilar or peritracheal lymphadenopathy, n=51; parenchymal lesion, n=27; pleural effusion, n=3; atelectasis, n=2). Chest x-rays demonstrated no evidence of compression or obstruction in six patients out of 10 with severe involvement documented by flexible bronchoscopy. No link was found between the radiological findings and the age of the patients.

Only 45 patients had a CT performed within the 15 day period before flexible bronchoscopy (range 0–15 days, mean 1 day). Five procedures (11.1%) required sedation. Eight patients (one with severe bronchoscopic involvement) were excluded from the analysis because the CT was performed outside the study period. No significant clinical or bronchoscopic difference was found between these eight patients and the 45 remaining. The CT abnormalities are described in table 2. It is noteworthy that bronchoscopy and CT showed exactly the same localisation of abnormalities. Hilar or peritracheal lymphadenopathy was identified in all patients on CT and in 43 on x-ray (95.5%). Nineteen patients out of 45 (42.2%) had a tracheobronchial luminal narrowing on CT (table 3). In this population, the chest x-ray under-diagnosed luminal narrowing (7/19) and to a lesser extent parenchymal lesions (34/40) compared to CT (table 2). The association between CT tracheobronchial luminal narrowing and severe bronchoscopic involvement was highly significant (p<0.001). CT sensitivity to predict severe bronchoscopic involvement was 100% (95% CI 74% to 100%). As a CT abnormality was found in 10 patients without severe bronchoscopic disease (six with an extrinsic compression <50% or an endoluminal mass <25%, and four with a normal flexible bronchoscopy), CT specificity was only 72.2% (95% CI 65.7% to 72.2%). On the other hand, the negative predictive value of CT was 100% (95% CI 91 to 100%). Based on CT results, a normal flexible bronchoscopy could have been avoided in 26 of these 45 children (57.7%).

Table 2

Radiological and CT findings in 45 children with active tuberculosis

Table 3

Correlation of bronchoscopic and CT findings in 45 children with active tuberculosis (p<0.001)


The use of flexible bronchoscopy for investigation has led to significant insights into various pulmonary disease processes, but for a surprising number of clinical situations there are either a lack of prospective data or a lack of consensus on the utility of airway endoscopy.13 For our patients with pulmonary tuberculosis, the use of routine flexible bronchoscopy permitted the diagnosis of bronchoscopic abnormalities in 49% of cases, guided steroid therapy in 22.6% and the identification of M tuberculosis in only 5.8%. No link has been found between bronchoscopic involvement and clinical features or chest x-ray findings. Conversely, the presence of a tracheobronchial luminal narrowing on the initial CT is statistically related to the severity of endobronchial involvement.

In a multicentre study where paediatric bronchoscopies were performed, pulmonary tuberculosis accounted for 6.2% of flexible bronchoscopy indications.11 Our results confirm the higher frequency of bronchoscopic abnormalities in children with pulmonary tuberculosis compared to adults.16 17 21,,23 Patients with severe bronchoscopic disease were all treated with oral steroids. Two other patients may have been treated excessively, highlighting the controversy over indications for steroid therapy. There is little information about the indications for and usefulness of steroids in the treatment of bronchial tuberculosis. In the 1960s, Nemir et al24 and Gerbeaux et al25 claimed that steroids were beneficial for the resolution of lesions, but did not provide data on long-term sequelae. In a more recent study performed on 29 patients with extrinsic bronchial compression of more than 50%, a significant difference was observed in the improvement of a bronchoscopic score when patients were treated with steroids.26 Others agree on the significance of rapid anti-tuberculosis treatment rather than steroid therapy to prevent bronchial complications.27 28 Bronchoscopic monitoring of patients treated with steroids allowed dose reduction. Serial bronchoscopy has confirmed regression of lesions, allowed decisions on extraction of granulomas by rigid bronchoscopy and revealed a case of partial bronchial stenosis. However, evolution of the disease seems positive among our population without the need for surgical intervention. In a study on 54 patients with pulmonary tuberculosis and treated with therapy equivalent to ours, De Blic et al described three resections by rigid bronchoscopy and two surgical procedures.17 However, because of the relatively brief follow-up of our patients, we are unable to exclude other sequela such as bronchiectasis or future bronchial stenosis. Future studies might look at the long-term sequelae of intraluminal lesions found on CT and at the ability of steroids to resolve lesions and symptoms.

Unlike the situation in adults, flex ible bronchoscopy does not significantly contribute to M tuberculosis identification in children. The culture yield of flexible bronchoscopy was low in our series (5.8%), but this is in keeping with published data with results ranging from 3.8% to 13%.2 4 7 16 17 Recommendations for microbiological confirmation in children are based on three early morning gastric lavages.29 The yield of gastric lavage can be improved if it is performed immediately after flexible bronchoscopy.30 Sputum induction is likely to be a promising technique for young children. Zar et al have demonstrated that the bacteriological yield of a single induced sputum sample was equivalent to that of three samples of gastric lavage.31

Flex ible bronchoscopy is recom mended by numerous aut hor s for the initial management of childhood pulmonary tuberculosis. However this procedure is still relatively uncomfortable for patients and any benefit depends on careful evidence-based patient selection. Unfortunately, no clinical features are likely to select children with an abnormal bronchoscopy. Moreover, in our study, a third of children with severe bronchoscopic involvement were asymptomatic. Chan et al showed the low diagnostic value of chest x-ray, with flexible bronchoscopy revealing endobronchial disease in 10 of 30 patients with no radiological evidence of endobronchial tuberculosis.16 We confirm the very poor sensitivity of chest x-ray which detected severe bronchoscopic involvement in only 40% of our patients. On the other hand, the discovery of tracheobronchial luminal narrowing on CT performed before flexible bronchoscopy appears to be a relevant criterion for suspecting severe endobronchial disease which could lead to a change in treatment. The most interesting finding is undoubtedly the negative predictive value of CT of 100%. To the best of our knowledge, this is the first paediatric study which has investigated this correlation. However, because of the low number of patients our results have to be interpreted cautiously. Nevertheless, if the decision to perform flexible bronchoscopy had been based on CT results alone, about 60% (26/45) of the flexible bronchoscopies could have been avoided. The limitations of this protocol are the administration of radiation and the necessity for sedation or anaesthesia for some children. However in our experience, sedation or anaesthesia has more often been necessary for flexible bronchoscopy than for CT procedures (23% vs 11%). The rapid evolution of CT technology has allowed for lower radiation exposure and shorter acquisition times.32 33 Eventually the use of virtual CT bronchoscopy, which already has excellent correlation with bronchoscopy, could improve CT specificity for indicating flexible bronchoscopy.34,,36

In conclusion, flexible bronchoscopy remains a relevant tool in the diagnostic and therapeutic management of childhood pulmonary tuberculosis. However, the decision to perform flexible bronchoscopy should be guided by the possible therapeutic advantages of the procedure, which in our study only benefitted approximately 25% of patients. Thus we propose that initial indications for flexible bronchoscopy should be limited to patients where initial CT findings reveal tracheobronchial luminal narrowing. A prospective study should be conducted to validate our proposal and the benefits of routine CT.



  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the local ethics committee.

  • Provenance and peer review Not commissioned; not externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.